Recognition And Tracking Of Weld Seam Of Underwater Welding

The main content of the paper is that recognizing the paths of welding lines when robot welds under water, meanwhile, designing and encapsulating a sensor system of the robot when it works, and then a lot of weld experiments under water have been done to find out the best scheme. At last, a control system of fuzzy controller of the robot has been designed to prepare for the subsequence work and simulates the system with the software of Matlab, and the robot tracks weld seam real time under water with the fuzzy controller and analyses the effect.To design the sensor of the weld system, the main work is that CCD and laser and filter are encapsulated in an aluminum box, the laser is in front of the CCD, and they are in the same plane. To light weld seam, the laser must have an angle at 23.4° with the CCD, which can screen the best photos. And to weld under water, the aluminum box must be encapsulated with a glass box.Because there are lots of schemes can recognize weld seams, so the best one must be chosen from them. Through a lot of examinations, it is this one that smooth a welding seam image to filter waves first, and then give a threshold value to the upper image, then check the brim of the image with Krisch operator to distill it's brim, after these, calculate the average of the image's pels to get the center line of the welding seam, at last, use the method of the least average to get the center of the welding seam. The value of the center of the pels will be shown by the program with VisualC++6.0 after finishing the work, and carry the value to the control system of the welding torch to decide how it move next time. The scheme is effective through a lot of tests including the water is calmness and agitation when welding.In the paper fuzzy control has been used to control the weld system because the parameter of the traditional control of PID is very difficult to adjust, and use it to have a try to track the weld seam under water simply.